Background
Methylation of gene promoters is a mechanism by which tumour suppressor genes can be inactivated. The role of promoter methylation in carcinogenesis has been convincingly demonstrated when gene methylation constitutes one of two events causing inactivation of well-documented tumour suppressor genes. Examples include, familial stomach cancer in which the non-mutated allele of
CDH1 is silenced by promoter methylation [
1] and sporadic renal cell cancer and retinoblastoma in which the non-deleted alleles of
VHL and
RB respectively are silenced [
2,
3].
Distinction of genes whose methylation is causally associated with malignant transformation from those that are affected by non-specific methylation remains problematic. It is plausible that genes that are densely methylated in all cells within the leukaemic clone are more likely to be involved in tumourigenesis than those that are partially methylated in a low proportion of leukaemic cells. Also genes that are methylated in a high proportion of cases seem more likely to be pathogenically important. Parallel evidence of gene silencing and evidence that the affected gene is a tumour suppressor gene greatly strengthens the case for a causal role in tumourigenesis.
Promoter methylation can occur as a non-specific "bystander" event affecting genes that are already silent in non-malignant tissue. For example, Keshet
et al. reported that of 106 genes whose promoters were methylated in colon cancer cell lines, 91 were already inactive in normal colon [
4]. Similarly, in acute lymphoblastic leukaemia (ALL), the methylated gene
TIMP3 was not expressed regardless of its methylation status [
5].
Gene promoter methylation has been reported for an, as yet, small number of genes in ALL [
5‐
9]. Genes that are reported to be methylated in ALL are involved in many cellular processes including growth regulation, apoptosis, cell adhesion, and others [
6,
8,
9] and therefore gene silencing by methylation is hypothesised to be an important contributor to leukaemogenesis. An example of candidate epigenetic silencing in the initiation and progression of leukaemogenesis involves
CDKN2B. This region frequently undergoes loss of heterozygosity (LOH) in ALL [
10,
11]. Dynamic changes of
CDKN2B promoter methylation have been reported during human myeloid development [
12] suggestive of a role in normal haematopoiesis.
CDKN2B promoter methylation, detected by methylation specific PCR, has been repeatedly reported in ALL leading to claims that this methylation in involved in leukaemogenesis [
13‐
16]. However,
CDKN2B methylation was neither dense, clonal nor prevalent in the reported cases.
For most of the reported gene methylation events in ALL, the proportion of affected cells and the density of methylation have not been quantified. For example, positivity in a methylation specific PCR assay indicates the presence of methylated alleles, but not their relative proportions. Array based methods, although not quantitative, have also been used to screen for methylated genes [
7].
Methylation-specific multiplex ligation-dependent probe amplification (MS-MLPA), a modification of MLPA, was developed as a tool for quantifying methylation at CpG sites located at methylation-sensitive restriction sites, by including a digestion step with a methylation-sensitive restriction enzyme such as HhaI [
17]. MLPA can also be used to measure gene dose. Therefore, MS-MLPA allows the rapid, simultaneous analysis of both copy number and methylation at a number of gene promoters. Here we use MS-MLPA to quantify methylation of candidate tumour suppressor genes in paediatric ALL.
Discussion
This study demonstrates that the
TES promoter is densely methylated in a high proportion of childhood ALL. Eighteen of twenty tested leukaemia bone marrow aspirate samples were densely methylated at the
TES promoter, whereas matched remission marrow, normal peripheral blood and bone marrow samples were unmethylated. In addition, eight of nine B-lineage ALL xenograft samples showed dense methylation of the
TES promoter. The proportion of ALL cases that showed methylation of
TES is among the highest for any genes reported to date [
6] and similar to levels reported by Hesson
et al. [
24] and Taylor
et al. [
7]. Taylor
et al. found methylation in
DCC in 9 of 10 cases of precursor B ALL and in
RUNDC3B,
KCNK2, and
DLC1 in 7, 7 and 8 (respectively) of these 10 cases [
7]. And in a recent study, Hesson
et al. reported
RASSF6 promoter methylation in 48 of 51 B ALL and 12 of 29 T ALL cases [
24]. For comparison, some of the genes that are commonly referred to as methylated in ALL,
e.g. the
p15 promoter have shown methylation of only 18% of alleles [
6].
TES is located in 7q31.2, a region showing frequent loss of heterozygosity in myeloid malignancies [
25] (between D7S2554 and D7S2460). In addition, loss of heterozygosity at 7q31 occurs in gastric cancer [
26], prostate cancer [
27], breast cancer [
28] and others (see Tobias
et al. for an overview [
19]). The frequent LOH implies the presence of at least one tumour suppressor gene, although the absence of mutations in candidate genes has led to suggestions that regulatory gene(s) within the region might be inactivated by epigenetic mechanisms [
29,
30]
TES is a putative tumour suppressor gene. Drusco
et al. concluded that
TES acted as a tumour suppressor gene
in vivo, given that
TES knockout mice showed an increased susceptibility to carcinogen (nitrosomethylbenzylamine) - induced gastric cancer [
31]. In addition, restoration of
TES by adenoviral transduction of non-
TES expressing breast cancer and uterine sarcoma cell lines inhibited their growth by induction of apoptosis [
32]. Additionally the tumourigenic potential of these transduced cell lines was significantly reduced in nude mice. Furthermore, forced
TES expression in a non-expressing, invasive ductal breast carcinoma cell line had an inhibitory effect on proliferation, on anchorage-independent growth in agarose and on colony forming ability [
33].
Available evidence suggests that apart from deletion, the commonest mechanism of
TES inactivation is epigenetic.
TES methylation has been shown in primary tumours, including glioblastomas (18 of 31) [
20] and ovarian cancer [
19]. Methylation at a single site in the
TES promoter has been reported for several cell lines including lymphoid leukaemia, breast cancer and pancreatic cancer cells [
18].
TES methylation is closely associated with loss of
TES expression in cell lines [
18] and in glioblastoma cells [
20]. By using xenograft and immortalised leukaemia cell lines, we have now directly demonstrated the reciprocal relationship between methylation and expression.
In contrast to epigenetic inactivation,
TES mutations have been reported in only three cell lines [
18,
19]. We investigated whether ALL5 and ALL19 (both partially methylated) harboured mutations; however
TES coding mutations were not detected by exon sequencing.
We confirmed that there is substantial down-regulation of
TES expression in an independent cohort of ALL cases and in B ALL xenografts. The marked down-regulation of
TES in virtually all cases of ALL, compared to normal precursor cells, indicates that
TES methylation suppresses expression that is present in relevant precursor cells. Although
TES expression levels were not confirmed in these ALL cases, we had previously demonstrated excellent correlation between these array results and qRT-PCR measurements for the majority (33/48) of selected genes [
34]. Additionally, our array results show consistent down-regulation with both
TES - specific probes used. Importantly another published series of 87 cases of B-lineage ALL showed that
TES was downregulated compared to normal bone marrow and normal haematopoietic cells, being the second most highly ranked down-regulated gene [
35]. Furthermore, within the cohort of Ross
et al.,
TES showed substantial down-regulation in B-lineage ALL compared to MLL-translocation ALL, or T-cell ALL [
36].
TES, a highly conserved protein (Additional file
3, Figure S3), is composed of three C-terminal LIM domains and a PET (prickle, espinas and testin) domain of unknown function. LIM domains are 50-60 amino acids in size and are believed to be involved in protein-protein interactions. LIM-domain containing proteins are classified into 4 groups; groups 2, 3 and 4 being predominantly localised to cytoskeleton-associated structures including focal adhesion complexes, whereas group 1 LIM proteins are predominantly nuclear.
TES, a group 3 LIM domain protein, is a component of the focal adhesion complex and localises to cell-matrix adhesions, cell-cell contacts and to actin stress fibres. In mice, Tes has been shown to interact or colocalise with cytoskeletal proteins including actin, zyxin, Mena, VASP, talin, α-actinin, and paxillin [
37]. Tes recruitment to the focal adhesion complex appears to be mediated by zyxin through the LIM1 domain [
33]. Tes also binds Mena, which inhibits Mena's ability to interact with FPPPP-motif proteins, such as zyxin, thus displacing Mena from focal adhesion complexes [
38]. Over-expression of TES leads to loss of Mena from focal adhesions, increased cell spreading and decreased cell motility [
37‐
40]. This suggests that Tes downregulates Mena-dependent cell motility, implying that loss of Tes might enhance cell mobility [
38]. As haematopoietic development requires coordinated bone marrow retention, adhesion and cell migration [
41], we suggest that silencing of
TES might contribute to ALL by interfering with normal interactions and adhesion between progenitors and stroma, with increased motility of immature progenitors, resulting in premature release of progenitors from bone marrow niches.
Additionally, TES has been detected in the nucleus, specifically the nucleolus, and the endoplasmic reticulum [
42]. It is proposed that the nucleolar localisation involves an alternative, closed confirmation state in which the N-terminus binds to the third LIM domain of TES [
42]. Therefore TES, similar to many of the other cytoplasmic LIM proteins, may shuttle into, and have a functional role within the nucleus. The possibility of additional complexity in the functional roles of TES is raised by our identification of a previously unreported short transcript of
TES in normal cells. If translated, this truncated protein would be very similar to the LIM-less TES proteins designed by Coutts
et al. [
37]. These LIM-less TES proteins do not localise to focal adhesions but are found associated with actin stress fibres. The function of the LIM-less splice variant is unknown, but it may compete with full-length TES activity.
Interestingly, other LIM domain proteins have been implicated in leukaemogenesis. For example, the group 1 nuclear-localised LIM protein LMO2 acts as an oncogenic protein in T-cell ALL [
43]. The role of LMO2 in oncogenesis was also demonstrated in two separate gene therapy trials for X-linked SCID, which were halted when five (out of 19) patients developed leukaemia. Remarkably four of the patients had insertion of the therapeutic retrovirus upstream of the LMO2 locus with subsequent over-expression of the LMO2 gene [
44]. LMO2 appears to act as a bridging molecule to assemble haematopoietic transactivating complexes and is essential for development of haematopoietic lineages [
45].
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
RJW and IMM designed the research and co-wrote the manuscript. RJW performed all the experiments, except for the microarray expression experiments (URK), and RJW, URK, SS and IMM performed data analysis. All authors read and approved the final manuscript.